bioRxiv preprint doi: https://doi.org/10.1101/2020.09.04.282962; this version posted September 5, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 1 The Pleistocene species pump past its prime: 2 evidence from European butterfly sister species 1 1 2 3 Sam Ebdon* , Dominik R. Laetsch , Leonardo Dapporto , 3 4 5 4 Alexander Hayward , Michael G. Ritchie , Vlad Dinc˘a , Roger 6 1 5 Vila , and Konrad Lohse 1 6 Institute of Evolutionary Biology, University of Edinburgh, 7 Edinburgh, EH9 3FL, UK 2 8 ZEN lab, Dipartimento di Biologia dell'Universit`adi Firenze, 9 Firenze, Italy 3 10 Centre for Ecology and Conservation, University of Exeter, 11 Penryn Campus, Cornwall, TR10 9FE, UK 4 12 Centre for Biological Diversity, School of Biology, University of St 13 Andrews, Fife KY16 9TH, UK 5 14 Ecology and Genetics Research Unit, University of Oulu, Oulu, 15 Finland 6 16 Institut de Biologia Evolutiva (CSIC - Universitat Pompeu 17 Fabra), Passeig Mar´ıtimde la Barceloneta 37, ESP-08003 18 Barcelona, Spain 1 bioRxiv preprint doi: https://doi.org/10.1101/2020.09.04.282962; this version posted September 5, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 19 * correspondence to [email protected] 20 21 Abstract 22 The Pleistocene glacial cycles had a profound impact on the ranges and genetic 23 make-up of organisms. Whilst it is clear that the current contact zones between 24 sister taxa are secondary and have formed during the last interglacial, it is un- 25 clear when the taxa involved began to diverge. Previous estimates are unreliable 26 given the stochasticity of genetic drift and the contrasting effects of incomplete 27 lineage sorting and gene flow on gene divergence. We use genome-wide tran- 28 scriptome data to estimate divergence for 18 sister species pairs of European 29 butterflies showing either sympatric or contact zone distributions. We find that 30 in most cases species divergence was initiated before the Pleistocene, substan- 31 tially earlier than assumed previously, and that post divergence gene flow is 32 restricted to contact zone pairs, although they are not systematically younger 33 than sympatric pairs. This suggests that contact zones are not limited to early 34 stages in the speciation process, but can involve notably old taxa. 35 Introduction 36 Divergence in allopatry provides a simple null model of speciation [1]. Fol- 37 lowing geographic isolation and given enough time, reproductive isolation is 38 inevitable as incompatibilities will eventually become fixed as a result of genetic 39 drift and/or selection [2{4]. Taxa that evolved partial reproductive isolation 40 in allopatry may come into secondary contact as a result of range shifts and { 41 depending on their degree of reproductive isolation and niche overlap { either 42 form a contact zone or invade each other's range [5, 6]. If allopatric divergence 2 bioRxiv preprint doi: https://doi.org/10.1101/2020.09.04.282962; this version posted September 5, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 43 dominates speciation, then local alpha diversity for a given clade cannot accrue 44 until secondary sympatry is achieved [7]. Thus the forces that facilitate or ham- 45 per secondary sympatry and the timescale over which this occurs have profound 46 consequences both for speciation and the spatial distribution of species diversity. 47 While modern ranges only provide a snapshot of the dynamic history of range 48 shifts, understanding the extent to which current range overlap between closely 49 related species can be explained by their speciation history and vice versa has 50 been at the core of speciation research [8]. 51 The glacial cycles of the Pleistocene had a profound effect on current diver- 52 sity of temperate ecosystems [9{11]. Populations of temperate taxa in Europe 53 were isolated in ice-free refugia around the Mediterranean basin (Iberia, Italy, 54 the Balkans and the larger Mediterranean islands) as glaciers encroached. The 55 observation that the geographic ranges of many young taxa are restricted to in- 56 dividual glacial refugia in southern Europe [9, 12{14] suggests that this repeated 57 separation into and expansion out of glacial refugia has played a major role in 58 their origin. The availability of allozyme and mitochondrial (mt) data in the 59 80s and 90s has spurred an abundance of case studies on intra- and interspecific 60 diversity of European taxa including detailed investigations of hybrid zones in 61 taxa ranging from fire-bellied toads [15], the house mouse [16], grasshoppers 62 [17, 18] to plants [19] and marine mussles [20]. The pervading evidence from 63 these studies is that genetic diversity within and in, many cases, divergence 64 between species is structured by refugia [9, 21, 22]. 65 When was divergence between sister species initiated? 66 While it is clear that the hybrid zones we observe today are secondary contacts 67 that formed after the last glacial maximum and may have formed many times 68 over throughout the Pleistocene, it is far from clear when divergence between 3 bioRxiv preprint doi: https://doi.org/10.1101/2020.09.04.282962; this version posted September 5, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 69 the sister taxa involved was initiated. One possibility is that the Pleistocene 70 glacial cycles initiated species divergence directly by separating populations into 71 allopatric refugia (i.e. a 'species pump' [23]). Another possibility is that the 72 initial divergence between sister species predates the Pleistocene, and so, any 73 build-up of reproductive isolation during the Pleistocene (e.g. via the fixation of 74 intrinsic incompatibilities and/or reinforcement) occurred in populations that 75 were already partially diverged. If the Pleistocene species pump hypothesis 76 is correct, we would expect species divergence times across sister pairs to be 77 concentrated during or at the beginning of the Pleistocene about 2.6 million 78 years ago (MYA). The idea that Pleistocene divergence acted as a species pump 79 was first proposed in the context of American faunas [23{25], but has also 80 dominated phylogeographic studies on European sister taxa [e.g. 9, 26{29]. 81 Other studies including some of the early work on European contact zones [5, 17] 82 conclude that initial divergence of the taxa involved may substantially predate 83 the Pleistocene [9, 30{32]. An important question to resolve, then, is whether 84 divergence of such sister taxa is the result of a 'Pleistocene species pump' or has 85 an older, deeper origin? 86 A corollary for the hypothesis of allopatric speciation in different refugia 87 is that range overlap is secondary. Since species can more easily invade each 88 others ranges once sufficient premating barriers and ecological differentiation 89 have developed, we would expect species pairs with overlapping ranges to be 90 older overall than those without range overlap, all else being equal [8]. Support 91 for this prediction comes from comparative studies showing that the proportion 92 of range overlap (degree of sympatry [33]) is positively (albeit weakly) corre- 93 lated with genetic divergence [6, 34]. However, a recent study in Chorthippus 94 grasshoppers shows that subspecies that hybridise across contact zones can be 95 older than currently sympatric species [35]. 4 bioRxiv preprint doi: https://doi.org/10.1101/2020.09.04.282962; this version posted September 5, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 96 Mito-nuclear discordance 97 Age estimates for recently diverged taxa have largely relied on single locus phy- 98 logenies and ignored incomplete lineage sorting. Hewitt [14] summarises age 99 estimates for European hybrid zones taxa including mammals, insects, amphib- 100 ians, and reptiles, which range from hundreds of thousands to several million 101 years ago. However, given that these estimates are based on different mark- 102 ers and calibrations, the extent to which glacial cycles have initiated speciation 103 events remains unknown. Estimates based on mitochondrial (mt) data are par- 104 ticularly unreliable for at least three reasons. First, the mutation rate of mtDNA 105 is highly erratic [36]. Second, given the stochasticity of coalescence, the ances- 106 try of a single locus (however well resolved) is a very poor measure of species 107 divergence. In the absence of gene-flow divergence at a single locus may sub- 108 stantially predate the onset of species divergence, while it may be much more 109 recent in the presence of gene flow [37, 38]. Mito-nuclear discordance in both 110 directions has been found in a large number of animal systems [39] including 111 several closely related species of European butterflies [40{42].